Improved Scaling Laws in Linear Regression via Data Reuse
원문 초록
Neural scaling laws suggest that the test error of large language models trained online decreases polynomially as the model size and data size increase. However, such scaling can be unsustainable when running out of new data. In this work, we show that data reuse can improve existing scaling laws in linear regression. Specifically, we derive sharp test error bounds on $M$-dimensional linear models trained by multi-pass stochastic gradient descent (multi-pass SGD) on $N$ data with sketched features. Assuming that the data covariance has a power-law spectrum of degree $a$, and that the true parameter follows a prior with an aligned power-law spectrum of degree $b-a$ (with $a > b > 1$), we show that multi-pass SGD achieves a test error of $\Theta(M^{1-b} + L^{(1-b)/a})$, where $L \lesssim N^{a/b}$ is the number of iterations. In the same setting, one-pass SGD only attains a test error of $\Theta(M^{1-b} + N^{(1-b)/a})$ (see e.g., Lin et al., 2024). This suggests an improved scaling law via data reuse (i.e., choosing $L>N$) in data-constrained regimes. Numerical simulations are also provided to verify our theoretical findings.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Near-Polynomially Competitive Active Logistic Regression
Active Learning for Manifold Gaussian Process Regression
Vector Contrastive Learning For Pixel-Wise Pretraining In Medical Vision
GRASP: Grouped Regression with Adaptive Shrinkage Priors
CAWR: Corruption-Averse Advantage-Weighted Regression for Robust Policy Optimization
Scaling-Up the Pretraining of the Earth Observation Foundation Model PhilEO to the MajorTOM Dataset
토론
이 논문과 연결된 포럼 글이 아직 없습니다.